Stability and uncertainty assessment of geoelectrical resistivity model parameters: a new hybrid metaheuristic algorithm and posterior probability density function approach

Abstract

Estimating a reliable subsurface resistivity structure using conventional techniques is challenging due to the nonlinear nature of the inverse problems. The performance of the inversion techniques can be pretty ambiguous based on the optimal error, although traditional methods have proven to be quite effective. In this work, the impacts of the constraints accessible from a borehole are examined for further assessment and to enhance algorithm effectivity. The vPSOGWO strategy is a new approach that is based on a model search space without any prior information, and it describes the hybridization of particle swarm optimization (PSO) with the Grey Wolf Optimizer (GWO). To understand the efficiency and novelty of the algorithm, it has been validated on two different kinds of synthetic resistivity data with various sets of noise and, subsequently, applied to three field datasets of different geological terrains. The analyzed results suggest that the subsurface resistivity model shows considerable uncertainty. Thus, it is superior to examine the histograms and posterior probability density functions (PDFs) of such solutions to exemplify the global solution. A PDF with a 68.27 % confidence interval (CI) selects a region with a higher probability. Therefore, the inverted models are used to estimate the mean global solution and the most negligible uncertainties, where the mean global solution represents the best solution. Our vPSOGWO-inverted outcomes have been proven to be more accurate than classic PSO, GWO, and state-of-the-art variants of classic approaches. As a result, this novel method plays a vital role in vertical electrical sounding (VES) data inversion.